How to industrialize the use of Artificial Intelligence (AI)

The 3 phases of the industrialization of (generative) artificial intelligence in the corporate environment

Prof. Dr. Martin Przewloka, November 2023

Introduction:

A year has passed: The hype - if that's what it is - surrounding Generative Artificial Intelligence, represented in particular by OpenAI, Aleph Alpha, Google Bard and others, is still present, although a little calm seems to have returned. In any case, one thing has happened in this relatively short period of time: Artificial intelligence (AI) is now no longer just a technology, but has become tangible and imaginable. We all currently recognize what machines called digital computers are obviously capable of, even if they work extremely inefficiently, but can and will ultimately take over valuable tasks from humans. The latter has long been depicted in visions such as films, but until recently was only perceived as entertainment and a topic of conversation under the heading of "science fiction".

As a result, there are hardly any companies today that do not think about the significance and adaptation of AI in their business area. "Where can I use this technology in my company? What is my competitive environment doing in this context? Am I already too late? Do I have the right employees in my company, or how do I get the right specialists?" These are just some of the questions that are being openly discussed today. Artificial intelligence is the driving force behind seminars, congresses, trade fairs and much more. And in just 5 days, you can currently become certified as an AI manager.

However, the key question is quite simple to formulate, as there is no longer any doubt that AI will find its way into almost all domains of our lives and, of course, into industry: "How should my company approach this technology, how should it develop and how can it benefit in the long term?"

?

The answer: Going through 3 evolutionary stages of AI industrialization with a focus on generative AI leads to long-term success

We have seen this time and again in the past. Especially in the context of technologies, the risk of overvaluation is very high. The first to do so will only rarely benefit from the time advantages that may exist, especially if the goals and expectations are too high and ambitious. However, short-term successes are essential for building trust and security in this domain and therefore also form the basis for strategic investments.

In the first stage, it is therefore the short-term successes that make the introduction of AI plausible. The key word here is efficiency gain and thus necessarily leads to the replacement of human labor by a machine. Nothing else underlies the different phases of the industrial revolution historically. The second stage requires the creation of a competitive position. The mere replacement of humans by machines cannot generate this advantage, as every competitor can or will act in the same way. Domain-specific knowledge will play a significant role and therefore still lead to machine-driven execution of previously human activities. However, the latter will ultimately be augmented or finalized by human expertise. Only in the third stage will humans and machines work collaboratively. Completely new things can emerge here, be it new products or services, but also completely new processes and methods of industrial value creation.

?

Stage 1: Generating efficiencies

A quick start requires quick successes. If these do not materialize, it will not be possible to develop a suitable AI vision for the company, as there is a lack of positive experience. Visionary promises, which are also disseminated by the media via all digital channels, can lead to ill-considered action and thus to disappointment. In this respect, it makes sense to align the first projects closely with the current business area and to optimize the associated processes and activities. Replacing human activities with automated systems is a good idea in precisely those areas where time-consuming, often routine-driven processes predominate.

The following key points indicate which actions should be implemented in this stage.

• Identify appropriate use cases (increase efficiencies). Points of reference are time-intensive processes, routine processes, clearly definable activities that are to be carried out by a machine application in future.
• Identify potential risks / develop mitigations. This applies not only to the implementation of use cases, but also to their operation combined with flexibility in dealing with any problems that arise.
• Use existing, commercial or available AI applications, i.e. avoid costly in-house and special developments for the time being.
• Creation of a governance model and its gradual refinement. Create trust for the AI-led application areas by means of suitable communication. Avoid possible organizational changes in the company through the use of AI.
• Identify initial points of reference for safe, responsible and trustworthy AI.
• Develop business cases and use them to create an initial AI roadmap 1.0. If necessary, define initial KPIs and check their suitability.

The possible use cases in connection with generative AI at this stage are strongly oriented towards currently available applications. These include, in particular, systems that can be subsumed under large language models (such as chatbots, text generators), as well as AI-based automation applications (such as automated processing of documents, analysis of unstructured information, identification of anomalies).

An initial takeover of existing, person-centered activities by machine applications will not go uncritically and smoothly within the company. It is therefore important to counter these potential risks at an early stage with open, trust-building, dialog-based communication. Ultimately, as with every introduction of a new technology (for example, the introduction of ERP systems in the past), there will have to be changes. The direct and indirect involvement of all employees in this change process should therefore already take place as part of this first stage.

?

Stage 2: Creating competitive advantages through the targeted use of AI

The next stage involves a much more strategic approach, i.e. the formation of a perspective that goes beyond short-term successes. This also involves making initial investment decisions with a medium-term horizon. The competition is sounded out here, be it through benchmarking or to identify unique selling points for future entrepreneurial activity.

Accordingly, this stage can be described using the following key points:

• Refinement and further development of the initial use cases with the aim of adapting them to the company's needs. The corresponding business cases must be created. In addition to further leveraging efficiency potential, these will include initial extensions to the current business models, e.g. by adding new services to the existing portfolio,
• Identification of domain-specific unique selling points that create competitive advantages and can be leveraged using specific AI applications.
• Identification of domain-specific unique selling points that create competitive advantages and can be leveraged using specific AI applications.
• Identifying and shaping the interfaces between man and machine for the secure integration of AI-dominated process steps into the company's entire value chain.
• Refinement of the governance model, definition of an AI roadmap 2.0 and its communication as well as derivation of initial organizational measures for implementation. Establishing change management at an early stage.
• Strategically driven creation of a technological infrastructure and its protection (data, cloud, intellectual property).

Stage 3: The AI-based transformation of the company through to leapfrog innovation

The aim of an AI-based transformation will be to expand the existing business model, whether by opening up new markets, developing new products and/or services or even diversifying completely. Innovations are now more tangible than ever and "creating something new" with the help of generative technologies is becoming a lived part of entrepreneurial activity. This can only succeed if people and machines work together, which means that the efficiency debate will initially take a back seat. The classic business case is only partially applicable, as the success of the new is only predictable on the basis of assumptions and therefore with risk. As a result, the company must transform itself. The portfolio transformation generated by AI alone is no longer sufficient. Interestingly, as is known from the history of industrialization, entrepreneurial skills with an affinity for risk will clearly emerge here, coupled with effects such as chance and luck.

There will be a significant change in the competitive environment: Fast-follower and late-follower strategies will find it difficult to hold their own. AI technology will make it possible for many to copy solutions that have been developed very quickly and without barriers - even if they are new and innovative - making it difficult to achieve the necessary differentiation in terms of an advantageous competitive position. In contrast, the leader will generate advantages by bringing together man and machine on a one-off basis, but will have to keep pushing forward the solutions developed in this way in order to maintain its own position.

The following key points describe the transformation stage:

• Developing an AI vision, adopting the strategic roadmap 3.0, securing the necessary investments and starting implementation.
• KPI-based project monitoring to measure the progress of the project and the subsequent iterative productive use.
• Consideration and creation of new values, such as responsible use of AI, ethical aspects, securing AI environments, etc.
• Development of own, new, highly specific AI models.
• Living digital transformation and implementing organizational changes. Add AI-related features to the corporate culture.
• Assume and defend a leading position.

Summary:

A step-by-step approach to implementing (generative) AI solutions in the corporate environment is necessary in order to successfully apply this currently highly valued technology. Initially, the rapid creation of experience with an appropriate short-term perspective is preferable to long-term, strategically plannable action. In this way, areas of success and problems can be identified at an early stage and transformational steps can be initiated on the basis of the relevant experience. Paradoxically, rudimentary human labor will initially be replaced by machine labor before the actual potential can be leveraged in the next step by merging humans and machines. It is only in this final stage that an awareness of the responsible use of AI will develop. In the current hype phase, this is obviously still falling by the wayside.